CS 640 1

The World Wide Web

Outline

Background

Structure

Protocols

CS 640 2

WWW Background• 1989-1990 – Tim Berners-Lee invents the World Wide

Web at CERN– Means for transferring text and graphics simultaneously

– Client/Server data transfer protocol• Communication via application level protocol

• System ran on top of standard networking infrastructure

– Text mark up language• Not invented by Bernes-Lee

• Simple and easy to use

• Requires a client application to render text/graphics

CS 640 3

WWW History contd.• 1994 – Mark Andreesen invents MOSAIC at National Center for

Super Computing Applications (NCSA)– First graphical browser– Internet’s first “killer app”– Freely distributed– Became Netscape Inc.

• 1995 (approx.) – Web traffic becomes dominant– Exponential growth– E-commerce– Web infrastructure companies– World Wide Web Consortium

• Reference: “Web Protocols and Practice”, Krishnamurthy and Rexford

CS 640 4

WWW Components• Structural Components

– Clients/browsers – to dominant implementations

– Servers – run on sophisticated hardware

– Caches – many interesting implementations

– Internet – the global infrastructure which facilitates data transfer

• Semantic Components– Hyper Text Transfer Protocol (HTTP)

– Hyper Text Markup Language (HTML)• eXtensible Markup Language (XML)

– Uniform Resource Identifiers (URIs)

CS 640 5

Quick Aside – Web server use

Source: Netcraft Server Survey, 2001

CS 640 6

WWW Structure• Clients use browser application to send URIs via HTTP to servers

requesting a Web page• Web pages constructed using HTML (or other markup language)

and consist of text, graphics, sounds plus embedded files• Servers (or caches) respond with requested Web page

– Or with error message

• Client’s browser renders Web page returned by server– Page is written using Hyper Text Markup Language (HTML)– Displaying text, graphics and sound in browser– Writing data as well

• The entire system runs over standard networking protocols (TCP/IP, DNS,…)

CS 640 7

Uniform Resource Identifiers

• Web resources need names/identifiers – Uniform Resource Identifiers (URIs)– Resource can reside anywhere on the Internet

• URIs are a somewhat abstract notion– A pointer to a resource to which request methods can be applied to

generate potentially different responses• A request method is eg. fetching or changing the object

• Instance: http://www.foo.com/index.html– Protocol, server, resource

• Most popular form of a URI is the Uniform Resource Locator (URL)– Differences between URI and URL are beyond scope– RFC 2396

CS 640 8

HTTP Basics• Protocol for client/server communication

– The heart of the Web– Very simple request/response protocol

• Client sends request message, server replies with response message

– Stateless– Relies on URI naming mechanism

• Three versions have been used– 09/1.0 – very close to Berners-Lee’s original

• RFC 1945 (original RFC is now expired)

– 1.1 – developed to enhance performance, caching, compression• RFC 2068

– 1.0 dominates today but 1.1 is catching up

CS 640 9

HTTP Request Messages• GET – retrieve document specified by URL• PUT – store specified document under given URL• HEAD – retrieve info. about document specified by URL• OPTIONS – retrieve information about available options• POST – give information (eg. annotation) to the server• DELETE – remove document specified by URL• TRACE – loopback request message• CONNECT – for use by caches

CS 640 10

HTTP Request Format

• First type of HTTP message: requests– Client browsers construct and send message

• Typical HTTP request:– GET http://www.cs.wisc.edu/index.html HTTP/1.0

request-line ( request request-URI HTTP-version)headers (0 or more)<blank line>body (only for POST request)

CS 640 11

HTTP Response Format

• Second type of HTTP message: response– Web servers construct and send response messages

• Typical HTTP response:– HTTP/1.0 301 Moved Permanently

Location: http://www.wisc.edu/cs/index.html

status-line (HTTP-version response-code response-phrase)headers (0 or more)<blank line>body

CS 640 12

HTTP Response Codes

• 1xx – Informational – request received, processing• 2xx – Success – action received, understood, accepted• 3xx – Redirection – further action necessary• 4xx – Client Error – bad syntax or cannot be fulfilled• 5xx – Server Error – server failed

CS 640 13

HTTP Headers

• Both requests and responses can contain a variable number of header fields– Consists of field name, colon, space, field value

– 17 possible header types divided into three categories• Request

• Response

• Body

• Example: Date: Friday, 27-Apr-01 13:30:01 GMT• Example: Content-length: 3001

CS 640 14

HTTP/1.0 Network Interaction• Clients make requests to port 80 on servers

– Uses DNS to resolve server name

• Clients make separate TCP connection for each URL– Some browsers open multiple TCP connections

• Netscape default = 4

• Server returns HTML page– Many types of servers with a variety of implementations

– Apache is the most widely used• Freely available in source form

• Client parses page– Requests embedded objects

CS 640 15

HTTP/1.1 Performance Enhancements• HTTP/1.0 is a “stop and wait” protocol

– Separate TCP connection for each file• Connect setup and tear down is incurred for each file• Inefficient use of packets• Server must maintain many connections in TIME_WAIT

• Mogul and Padmanabahn studied these issues in ’95– Resulted in HTTP/1.1 specification focused on performance

enhancements• Persistent connections• Pipelining• Enhanced caching options• Support for compression

CS 640 16

Persistent Connections and Pipelining• Persistent connections

– Use the same TCP connection(s) for transfer of multiple files

– Reduces packet traffic significantly

– May or may not increase performance from client perspective• Load on server increases

• Pipelining– Pack as much data into a packet as possible

– Requires length field(s) within header

– May or may not reduce packet traffic or increase performance• Page structure is critical

CS 640 17

HTML Basics• Hyper-Text Markup Language

– A subset of Standardized General Markup Language (SGML)– Facilitates a hyper-media environment

• Embedded links to other documents and applications

• Documents use elements to “mark up” or identify sections of text for different purposes or display characteristics

• Mark up elements are not seen by the user when page is displayed• Documents are rendered by browsers• NOTE: Not all documents in the Web are HTML!• Most people use WYSIWYG editors (MS Word) to generate

HTML

CS 640 18

HTML Example

<HTML><HEAD><TITLE> PB’s HomePage </TITLE></HEAD><BODY><CENTER><IMG SRC = “bad_picture.gif” ALT = “ “><BR></CENTER><P><CENTER><H1>UW Computer Science Department</H1></CENTER>Welcome to my goofy HomePage!…<A HREF = http://www.cs.wisc.edu/~pb/mydogs_page.html> Spot’s Page </A></BODY></HTML>